Decoding of Superimposed Traces Produced by Direct Sequencing of Heterozygous Indels Dmitriev, D.A. & Rakitov, R.A.

Transcription

1 Decoding of Superimposed Traces Produced by Direct Sequencing of Heterozygous Indels Dmitriev, D.A. & Rakitov, R.A. Illinois Natural History Survey, Institute of Natural Resource Sustainability, University of Illinois at Urbana-Champaign, 1816 S. Oak st., Champaign IL,

2 Problem of double peaks on chromatogram A pair of allelic sequences properly aligned (A), unaligned (B), translated into a consensus (C), and resulted chromatogram (D). The following applications could be used to call double paeks PHRED, KB Basecaller, Sequencher

3 Reasons for double peaks Direct sequencing of diploid alleles containing heterozygous insertions/deletions. (Mixed trace downstream of the indel is formed by two allelic traces superimposed onto each other with a phase shift). Sequencing of unrelated templates and alternative splicing. Single nucleotide polymorphisms (SNPs) or base calling errors due to low quality of chromatogram (individual double picks)

4 Solutions Discard as uninterpretable. Use new sequencing technologies, such as pyrosequencing, which works with single DNA molecules. Separating the templates prior to sequencing via cloning into a vector or selectively amplifying one allele using allele-specific primers. Computational methods to extract information from mixed traces.

5 Computational methods for decoding of mixed traces Subtracting a reference sequence: PolyPhred, STADEN, CodonCode Aligner, Mutation Surveyor, InSNP, PolyScan, AutoCSA, and Tenney, A.E at al. (2007) application to automatically resolving double traces aligning the mixed sequence to genomic database Using reverse sequence as a reference: SeqScape, Varian Reporter, Champuru. Extracting information from an individual mixed trace (without reference): Shift Detector, CodonCode Aligner Ver. 2, Indelligent.

6 Phase shift and two renderings of the same alignment. Optimality criterion. V=n - (#mismatches) - (#insertions) - (#inserted bases) Number of solutions = 2 n-1 x k max

7 Dynamic optimization algorithm

8 Multiple cooptimal solutions

9 Accuracy of decoding of simulated mixed fragments formed by a single 5bp phase shift (1000 runs for each point)

10 Accuracy of decoding of simulated mixed 100 bp fragments formed with an insertion of variable size in the middle (1000 runs for each point)

11 Two aligned solutions of the same mixed fragment representing the transition between two phase sifts ( Long vs. Short indel)

12 Validation with human traces We used 104 ( bp, with indels 5-30 bp) traces of 198 recorded by Bhangale et al. (2005) from NCBI Trace Archive as having heterozygous indel. Sequencer was used to call second paeks. After reconstruction, traces were aligned with best matching human sequnces in NCBI Trace Archive. 102 traces reconstructed with a single indel, two with two indels. 67 traces without errors, 31 with 1-2 errors, 6 with 3-7 errors. Half of the fragments reconstructed without ambiguities. Mean of 99.1±1.25% of bases per fragment decoded correctly and unambiguously (in the same conditions ShiftDetector decoded only 72.5±6.47% of bases). About 60% of reconstructed mean of 0.66 errors per fragment were due base calling errors, mostly in low-quality trace regions.

13 Conditions for reconstructions with % accuracy Homologous fragments resulted from indel mutation. Analyzed fragment is significantly larger than the indel (at least 10 times; in human 92.3% of indels are 1-10bp). Low divergence between mixed traces (<5%; for human noncoding DNA, the average divergence is <0.1%, fruit fly 1-2%, sea squirts 4.5%). Multiple indels, if present, are well spaced. Methods relies on base calling software.

14 Indelligent interface

15 Reconstruction results

16 Acknowledgments We are thankful to Saurabh Sinha for valuable discussion and suggestions Tushar Bhangale for providing information which enable us to obtain human traces for testing. Chris Dietrich for helpful comments and support. The work was partially supported by NSF grants.